May Your Drug Prices be Evergreen

Abstract

I. INTRODUCTION

The intellectual property system has a simple and intuitive design at its core. From the store of activities that should be free to all people, we remove some, for a limited time and a limited purpose, in the hopes that the pause will rebound to the benefit of all of society.² This conceptualization echoes basic Lockean theories on the formation of government, in which individuals emerge from perfect freedom in the state of nature, choosing to relinquish certain liberties (and only certain ones) for these individuals’ mutual benefit.³ One can wax poetic about the complicated pathways of the intellectual property system—the intricacies of state and federal powers,⁴ the delicate dance of biosimilars,⁵ the vastness of open source and open science,⁶ and the strange overlap of different protection regimes.⁷ Nevertheless, the basic concept of the US intellectual property system is quite simple: give inventors the possibility of garnering a return from their innovations, and they will invest in creating those innovations and in sharing the fruits of their labors with society.⁸

At first glance, one might think the intellectual property system eschews competition. After all, the system is designed to grant benefits that block competitors, giving the rights holder free reign in the market, a result that is decidedly non-competitive.⁹ That perspective, however, only skims the surface of the theoretical bases of intellectual property. The reality is far more nuanced and layered when one plunges the depths of the system’s design.

At a fundamental level, the intellectual property system exudes a deep faith in the power of competition. Competition may be held in abeyance, but those who receive the benefit of a patent or exclusivity must pay for that privilege by disclosing sufficient information such that competitors will be able to step into the market. And as the protection clock winds down, other inventors can use that disclosure, making preparations to enter the competitive field or jump ahead to the next generation.¹⁰

Nowhere does this concept apply more fully than with pharmaceutical development. The processes of developing new drugs, conducting the clinical trials, obtaining FDA approval, and bringing the drugs to market are extraordinarily expensive. Scholars and commentators disagree over the magnitude of the cost,¹¹ but no matter how one measures it, big is big. The prospect that a second-comer could simply copy the drug after all that effort would deter even the heartiest of souls, and thus the intellectual property system provides the opportunity to secure a return. In idealized form, a company invests in developing a drug: when the company succeeds, it obtains market exclusivity for a period of time, and when the exclusivity expires, generic companies step in to create a vigorous competitive environment.

In discussing the pharmaceutical industry, the broader term ‘intellectual property’ should be used, rather than the narrower term ‘patent’. Although patent protection is a critical component of the incentive structure society provides for pharmaceutical development, it is not the only component.¹² The federal government offers more than 10 other forms of exclusivity that can be used to keep competitors at bay. Companies can earn exclusivity benefits for activities such as development of drugs for smaller populations¹³ or for conducting pediatric studies.¹⁴

Whether society grants intellectual property in the form of a patent or a regulatory exclusivity, the systems are designed such that after a period of time, competitors may enter. Information revealed in the patent allows others to create a competing drug (rather than going through the research again, themselves).¹⁵ Data used in clinical trials for the drug are available, after a period of exclusivity, so that follow-on generics need only prove their drug is the same, rather than repeating the original company’s safety and efficacy trials.¹⁶ Thus, when the benefit expires, competitors should step in and competition should drive prices down to competitive levels—at least in theory. The reality for pharmaceutical products, however, lies far from the system’s theoretical design.

It is no exaggeration to say that drug prices have skyrocketed. The cost of prescription medication is growing faster than any other form of health care spending, including hospitalization or nursing home care.¹⁷ These price increases can be seen in specialty drugs—such as the antimalarial drug Daraprim,¹⁸ which Martin Shkreli’s company famously increased from $13.50 per tablet to $750 per tablet—and in more common drugs—such as the rheumatoid arthritis drug Humira, whose price has increased by 126%.¹⁹

Inspired by the rise, new medications are entering the market at astoundingly high prices. Gilead’s new treatment for hepatitis C lists at a hefty $84,000 to treat a single patient,²⁰ Marathon’s muscular dystrophy drug debuted in the United States this year at $89,000, for a drug that reportedly can be obtained in other countries at $1500.²¹Prices like this can be cost-prohibitive. For example, the cost for the Department of Defense to treat all infected patients in the VA with Gilead’s hepatitis C drug, Sovaldi would amount to $12 billion—more than ‘20% of the department’s $57 billion medical budget in fiscal year 2014’.²² In California, just treating 3624 patients cost the state more than $387 million.²³ Put simply, states are forced to compromise, choosing between patient health and staying afloat.²⁴

The cost of drugs designated for small patient groups is particularly high. These are known as orphan drugs, and drug companies are rushing into the field. In fact, orphan drugs account for 40% of the new drugs approved in the United States. As one commentator has noted, in today’s pharmaceutical market, everyone seems to be an orphan.²⁵ A 2017 study found that the median price of a group of orphan drugs was $140,000 per patient, per year.²⁶ The price of ordinary drugs was nothing to sniff at, either. The median price for drugs outside the orphan category had climbed to almost $28,000 per patient, per year.²⁷

Europe has also faced rising drug prices amidst a pharmaceutical framework that provides a host of protections and exclusivities beyond traditional patent terms.²⁸The European Commission is currently conducting a review of the system, ‘assessing whether the pendulum has swung too far in favor of the pharma industry while potentially penalizing the generics sector, governments, and other payers and patients’.²⁹

In a competitive environment, other producers would enter the market, driving down these sky-high prices. Even in suboptimally competitive markets such as health care, one might expect to see some measure of competition, at least in certain circumstances. In particular, many drugs with high prices have been available far longer than the 20-year term of a patent, and the modern drug approval system is designed to encourage generic versions of drugs after that time. So why is the system failing?

Anecdotal evidence has identified strategic behaviors various companies have deployed to great effect. One such practice is ‘evergreening’, which can be defined as artificially extending the life of a patent or other exclusivity by obtaining additional protections to extend the monopoly period.³⁰ Scholarly work, including our own, has documented these behaviors as examples have emerged in individual cases and in press reports.³¹ What has been missing from the literature, however, is a comprehensive empirical view.³² Just how pervasive are such behaviors? Is it simply a matter of certain bad actors, to whom everyone points repeatedly, or is the problem endemic to the industry? Only by answering this question can we contemplate the extent to which reforms are needed, as well as the extent to which strategic behavior to block generic competition may be contributing to rising drug prices. This study answers the questions.

Providing a robust empirical analysis was no easy task. Transparency is not in the industry’s interests, and companies have been known to go to great lengths to camouflage strategic behavior.³³ After all, a pharmaceutical company would be loath to let regulators and legislators know what it is up to, let alone competitors who might mimic the clever strategies. To accomplish our study, we turned to government sources, analysing more than a decade of data published by the US Food and Drug Administration (FDA). This involved extracting and analysing detailed information on as many as 11 different aspects of roughly 1800 drugs.

The task would have been sufficiently challenging if the information were readily available. It was not. The project required teasing information painstakingly out of each monthly and annual publication, many of which are no longer available from the government in any form. Moreover, the complexities of pharmaceutical regulation and approval require intricate analysis of the information disclosed by the government, when that information is disclosed at all. In all, our work required assembling and analysing over 160,000 individual cells of data, all entered by hand.

The results, however, were striking, and they show a startling departure from the classic conceptualization of intellectual property protection for pharmaceuticals. The data demonstrate that throughout the industry, companies create serial barriers to hold off the type of competitive entry that is fundamental to our innovative system.

Key results from our 2005 to 2015 study include the following:

• Rather than creating new medicines, pharmaceutical companies are recycling and repurposing old ones. In fact, 78% of the drugs associated with new patents in the FDA’s records were not new drugs coming on the market, but existing drugs.
• Adding new patents and exclusivities to extend the protection cliff is particularly pronounced among blockbuster drugs. Of the roughly 100 best-selling drugs, more than 70% had their protection extended at least once, with almost 50% having the protection cliff extended more than once.
• Looking at the full group, almost 40% of all drugs available on the market created additional market barriers by having patents or exclusivities added on to them.
• Once a company starts down this road, there is a tendency to keep returning to the well. Eighty per cent of those who added protections added more than one.
• Among those adding more than one barrier, some were serial offenders, with almost half adding 4 or more protections and some adding more than 20.
• The problem is growing across time. The number of drugs that had a patent added on to them almost doubled during the time period. The addition of certain other types of barriers increased at an even greater rate, with some tripling.³⁴

These results may easily understate the landscape. In designing the methodology, we repeatedly adopted a conservative approach, following the path that would point away from suggesting a competitive barrier. In addition, the pharmaceutical industry has developed techniques for erecting competitive barriers that do not involve obtaining additional patents and exclusivities, techniques that would not be captured by our analysis.³⁵ Finally, we could only quantify those behaviors of which we are aware. Much behavior in the pharmaceutical industry remains obscured, and we cannot measure what we cannot see.³⁶

Thus, for the first time in the literature, this study definitively shows that stifling competition is not limited to a few pharma bad apples. Rather, it is a common and pervasive problem endemic to the pharmaceutical industry. Although the end of life for a patent or exclusivity may be a traumatic event in the life of a pharmaceutical enterprise, companies increasingly decline to ‘go gentle into that good night’.³⁷

In short, this is not an image of innovation and competitive entry. It is an image of a system that provides for repeated creation of competition-free zones, pushing a competitive market further and further out into the future. The problem is not only pervasive and persistent, but it is also growing across time. Against this backdrop, it is no wonder that drug prices are skyrocketing.

II. BACKGROUND

II.A. A brief tour of the modern drug approval process

The following section provides brief highlights of the modern drug approval process, with a focus on aspects relevant to our study.³⁸ The modern system for drug approval in the United States is a long and arduous process. Companies wishing to bring an entirely new drug to market must develop the drug, determine how to manufacture it on a mass scale and in a way that is consistently stable, and prove to the FDA that the drug is safe and effective through rigorous clinical trials. Survivors of this marathon—at least those whose innovation is significant enough to earn a patent—are rewarded with the right to exclude others from making, using, or selling the drug.³⁹

The cost of obtaining a patent is miniscule compared to the hundreds of millions of dollars necessary to take a drug through clinical safety and efficacy trials.⁴⁰Moreover, companies try to plant their patent stake in the ground as soon as possible, to mark off their territory and keep others out. Given both of these realities, companies obtain many patents that are never developed into viable products, including many patents that sit idly on the shelf.

With patenting occurring early in the drug development cycle, some of the patent term will have expired before the drug gets to market. Estimates suggest that the average remaining patent period for a new drug is 12 years. Although far less than a term of 20 years from the time of a patent application, 12 years of exclusivity is a considerable reward, particularly for a blockbuster drug that will garner many billions of dollars a year in revenue.

One should note that even with patents outside the pharmaceutical space, companies will not necessarily enjoy a full 20 years of exclusivity on the market. It takes time to develop and market any product, as well as time to get through the patent office’s approval process. In addition, many products contain numerous patents, along with trade secrets and other knowhow, such that a single patent will not lead immediately to a marketable product.⁴¹ The lag time for drug development, however, is likely to exceed the lag time for many other products, even if the difference is not a full 8 years.

All good things must come to an end, however,⁴² and when the patent expires, the system is designed so that generic companies can immediately step into a pharmaceutical market and compete. The Hatch-Waxman system, along with the accompanying regulatory and judicial structure, provides the vehicle for rapid entry of generic drugs. ⁴³ Under Hatch-Waxman, generic hopefuls can clear the legal and regulatory hurdles ahead of time in order to hit the ground running.

A generic company would not have the potential for monopoly returns from excluding others from the market, given that the generic will have nothing new to patent. Thus, generic companies would lack the financial incentive to engage in lengthy and costly clinical trials. Nor would repeating those trials necessarily represent a good use of societal resources, considering that the brand-name company has already established the safety and efficacy of the chemical formula.⁴⁴ In light of these constraints, Hatch-Waxman allows generic companies to reference the safety and efficacy data from the brand-name company’s original drug application, which is known as a ‘New Drug Application’ or ‘NDA’ for short.⁴⁵ The generic company need only demonstrate bioequivalence.⁴⁶ In other words, the generic company does not need to show that the formula is safe and effective, only that its product is the same as the brand.

As part of keeping prices low, generic companies generally do not engage in extensive advertising, either to providers or directly to consumers. Rather, they depend on drug substitution laws that allow pharmacists to substitute a cheaper, generic version when a physician prescribes a medication.

In creating the Hatch-Waxman system, Congress recognized that the US Patent and Trademark Office (USPTO) unfortunately grants many patents of dubious quality. The problem is not surprising, given that on average, the patent office spends only 18 hours across a 2-year period examining a patent application.⁴⁷ This is painfully little time for patents, particularly pharmaceutical patents that may contain hundreds of claims. Although the number of patent examiners has doubled since 2005,⁴⁸ the number of patents approved each year has doubled as well, rising to over 300,000 new patents in the fiscal year ending August 2017.

Patents of questionable validity can improperly block competitors out of the market. In addition, a different problem occurs when a perfectly valid patent is applied inappropriately to a drug. For example, the FDA requires companies to submit any patents that relate to a drug within 30 days of the drug’s approval. Under the Hatch-Waxman system for approval of generics, there are repercussions for brand-name companies that do not file within the proper time limits.⁴⁹ The FDA does not scrutinize the company’s representations, however, but merely records whatever the company submits in what is known as the ‘Orange Book’. Thereafter, a competitor seeking approval of a generic version of the drug must battle every patent listed in the Orange Book in relation to the drug.⁵⁰ Thus, simply listing a patent in the Orange Book can operate to block or delay competition, even if that patent does not cover the drug.

To address the problem of invalid patents or patents invalidly applied, Hatch-Waxman provides an incentive for generics to engage in such battles with an established brand-name company. Specifically, the first generic to successfully challenge a drug patent or the application of that patent to its generic application will be the only generic allowed on the market for 6 months.⁵¹ During this 6-month period, a duopoly market will exist, in which the only players are the brand-name company and the first generic.⁵²

The introduction of generics is a shock to the system for a pharmaceutical company. Prices can drop as much as 20% when the first generic enters the market; with multiple generics, the prices may eventually drop by 80–85%.⁵³ As a result, drug companies have a powerful incentive to delay competitive entry for as long as possible. Even small delays can have a big impact on a company’s bottom line. A few months of delay can be worth hundreds of millions of dollars for blockbuster drugs, whose revenues reach billions of dollars a year.⁵⁴

It should come as no surprise that drug makers do all they can to soften the blow of losing market monopoly. Some strategies to mitigate the effect of falling off the protection cliff are relatively straightforward, such as raising prices on those drugs that are still protected. Other strategies involve what is known as ‘evergreening’.⁵⁵Although commentators use the term in slightly different contexts,⁵⁶ we will use its broadest connotation of trying to refresh one’s monopoly protection on a drug.

Simple techniques can involve obtaining new protections on existing drugs by filing for additional patents, sometimes on methods of producing or manufacturing the drugs or on other aspects. For example, in an empirical study of secondary pharmaceutical patents between 1985 and 2005, Kapczynski, Park & Sampat found that secondary patents—covering ancillary elements of a drug such as formulation or method of use, as opposed to the primary chemical compound—were highly common.⁵⁷ These supplementary formulation patents added an average of 6.5 years of patent life, and supplementary method of use patents added an average of 7.4 years of patent life.⁵⁸ Other work has determined that secondary patents are likely to be overturned by generic challengers, if the case is litigated to completion.⁵⁹

More complex evergreening strategies involve developing new formulations, dosage schedules, or combinations that can be used to obtain new patents. These can be combined with attempts to move the market to the slightly altered product, by advertising extensively, pressuring doctors to write prescriptions with terms such as ‘Dispense as Written’ or ‘Brand Medically Necessary’, or even withdrawing the old product from the market entirely. Using these techniques, brand-name companies try to prevent pharmacists from being able to fill a prescription with a generic.⁶⁰ At the very least, the brand-name company might be able to bifurcate the market, with some patients moving to the new version for which no generic is available.

Many of these evergreening strategies involve applying for new patents. Even if the patents are of questionable validity, the process of challenging them through Hatch-Waxman litigation is expensive and lengthy for a generic, again allowing years of additional profits for the brand-name company.⁶¹

If companies are able to develop new formulations, dosage schedules, combinations, and the like, in a way that justifies obtaining new patents or exclusivity protections, these companies not only minimize damage from tumbling off the cliff but may also be able to delay going over the edge in the first place. Our data suggest that this is occurring in a widespread manner throughout the industry.

Consider the dementia medication Namenda, a blockbuster drug that was scheduled to lose patent protection in 2015.⁶² The company launched a longer-acting version of the original drug product and began encouraging patients and doctors to switch to the patent-protected, longer-acting version in order to undermine generic competition.⁶³ In our dataset we recorded one version of Namenda, approved in October 2003, for which the protection cliff was extended once in 2009. There was another version of Namenda approved in April 2005, for which the protection cliff was extended twice: once in 2007 until 2014 and once in 2009 until 2015. Finally, Namenda XR was approved in 2010, for which the protection cliff was extended once in 2011 until 2029. Throughout this series of maneuvers, Namenda was able to extend its protection at least 24 years.

These are not the only strategies companies use to extend protection. As described above, the FDA takes the company’s word for whether a patent should be listed as applying to a particular drug. The same is true for the company’s description of what uses of the drug are covered by the patent’s claims. Specifically, the FDA requires that the drug company submit a short statement describing the approved use (or uses) claimed by the patent, which the FDA then assigns a number and lists in the Orange Book as a use code.⁶⁴ Scholars have demonstrated that brand-name companies often submit use codes that are overbroad or inaccurate in describing the actual content of the patent.⁶⁵

Given that the FDA does not read or construe patent claims, generics have little recourse for correcting incorrect use codes.⁶⁶ In 2012, the Supreme Court ruled in Caraco that generic companies can file statutory counterclaims to seek correction of inaccurate use codes,⁶⁷ but the approach requires entering into the extensive legal dance of submitting a Paragraph IV certification, attracting a lawsuit from the brand-name company claiming you have infringed, and then successfully defending against that infringement suit.⁶⁸

In response to continued concerns about use codes, new FDA regulations, that became effective at the end of 2016, have established the ‘Orange Book Patent Listing Dispute List’.⁶⁹ To dispute the accuracy of a use code under this newly implemented system, one may submit a ‘statement of dispute’ to the FDA, which the FDA will then send to the company whose use code is in dispute. That company must confirm the correctness of the use code, or otherwise withdraw or amend the patent information, and must also include a description explaining how the existing or amended use code is accurate. The process, however, has no teeth. The FDA will simply post information on these use code disputes online under the ‘Orange Book Patent Listing Dispute List’.

In addition, although there is some penalty for failing to list a patent in the Orange Book in a timely manner,⁷⁰ the same is not true for use codes. Once a patent has been submitted, the company can determine at any point in the life of the patent that the patent covers a new use.⁷¹

Obtaining new patents from the Patent Office and adding use codes to the Orange Book are not the only ways to extend one’s protection on a drug. There are more than 10 different exclusivities one can obtain from the FDA, all of which can create competition-free zones for a drug company for a specified period of time. Sometimes called regulatory exclusivities or regulatory property,⁷² these programs were approved by Congress during periods in which Congress passed legislation opposed by the pharmaceutical industry.⁷³ Drug companies can apply for the benefits for a variety of reasons, including performing pediatric testing, performing other new clinical studies, developing so-called ‘Orphan Drugs’, and developing drugs for tropical diseases.⁷⁴ These benefits can operate to extend protection by adding to the length of the patent term, creating a time period in which other companies are not permitted to receive approval to market the drug or to use existing safety and efficacy data, adding to the length of already existing non-patent exclusivities, or providing for combinations of these benefits.⁷⁵ Additional details on these exclusivities can be found in the Methodology section and in the Results section.

Of course, attempts to block out competitors may not always be successful. Competitors may be able to overturn or avoid weak patents, challenging them as either invalid or not infringed. Other behaviors can be used to avoid the patent deluge in limited circumstances, such as producing a drug with a label designating a narrow use while knowing that physicians are likely to suggest off-label use of the drug.⁷⁶ Even where such approaches are successful, however, the cost and waste of rent-seeking behavior, and behavior to counter such rent seeking, does not provide a model of an effectively functioning market.

In short, despite the quaint theory that competitors will enter after a pharmaceutical patent expires, the reality is quite different. Numerous strategies and opportunities exist that allow companies to extend their protection and prolong the period of market monopoly for their drugs. Such game-playing involving patents and exclusivities has been explored primarily from a theoretical standpoint and through case studies, with no comprehensive, quantitative examination of such strategies across the industry. Our study fills the gap.

III. METHODOLOGY⁷⁷

III.A. Overview

We sought to compile a large volume of FDA data that would allow us to examine the prevalence and specific contours of patent and exclusivity game-playing in an empirically rigorous manner. We hypothesized that the behavior of repeatedly adding patents and exclusivities would be detectable in a widespread manner across drug products, and that such behavior is increasing across time.

We used data published in the FDA’s ‘Orange Book’ to test our hypotheses.⁷⁸ Locating FDA data and converting it into a format conducive to analysis was a formidable task. Although each monthly supplement and annual edition across time contains a wealth of information, only the most recent edition is available from the FDA.⁷⁹ We were able to locate archived copies of the monthly and annual editions from another researcher to obtain our source data. From that data, we extracted all the patent and exclusivity information from the 11 years of Orange Books included in our study, examining each to determine detailed information on the nature of the addition or change, as detailed in Section III. B.ii. Enormous effort was required to gather the dataset and render it usable for empirical analysis. Consistent with our prior practices,⁸⁰ as well as our commitment to transparency and high ethical standards in data-driven academics,⁸¹ this dataset is publicly available. As pharmaceutical pricing gains focus in public policy debates, we hope the dataset will assist other researchers, regulators, and the general public in future investigations into the pharmaceutical industry.

III.B. Methodology details

III.B.i. Just what are the cumulative and annual editions of the Orange Book?

At the beginning of each year, the FDA publishes an ‘Annual Edition’ of the Orange Book, with information current up to the last day of the previous year. The Annual Edition lists all approved drugs, whether they are on the market as of that moment, had never been marketed, or have been discontinued from marketing. The patent and exclusivity section of the Annual Edition contains information on the active patents and exclusivities attached to approved drugs.

The FDA also publishes a ‘Cumulative Supplement’ every month of the year, containing new information received and processed since the publication of that year’s annual edition. The FDA explains in the Orange Book that the Agency aims to update the cumulative supplement ‘by the end of the following month’s second work week (e.g., November’s supplement will be updated by the end of the second full work week in December)’ and that patent and exclusivity information is ‘current to the date of publication’.⁸² This lag in the publication of the Orange Book leads to some minor imprecision in terms of the date on which the information was submitted and the date in which it is published.

Each cumulative supplement lists both the new patents and exclusivities that were added in that specific month, as well as the patents and exclusivities added in earlier cumulative supplements from that year. Certain lines in the patent and exclusivity section in each cumulative supplement are marked with a symbol indicating that the listing was added to the Orange Book that month and had not appeared in previous cumulative supplements of the Orange Book from that year.⁸³

III.B.ii. Compiling the patent and exclusivity data

The process of compiling data on patents and exclusivities added to drugs between 2005 and 2015 consisted of three general stages: (1) transferring all patent and exclusivity additions from the cumulative supplement for each month between January 2005 and December 2015 to a central dataset; (2) transferring all patent and exclusivity information from the 2005 annual edition of the Orange Book to the dataset, so that this information serves as a reference for analysing the additions after the 2005 annual edition; and (3) double checking all of the entries in our dataset to minimize the likelihood of human error.

III.B.iii. Transferring patent and exclusivity data from the cumulative supplements

The first step in our data-gathering process was to transfer all patents and exclusivities marked as new additions from each month between January 2005 and December 2015 to a comprehensive dataset that included a wide range of information. For each patent or exclusivity, we recorded the active ingredient name, the product name, the New Drug Application (NDA) number, the month and year of the addition, whether the addition was a patent or exclusivity, the patent number (if applicable), the code(s) attached to the patent or the exclusivity code, the expiration date, the strength(s) of the drug to which the Orange Book addition applied, and whether a ‘delist request’ flag was attached to the patent.⁸⁴ After transferring the above information available in the Orange Book, we used the Drugs@FDA database—an online repository of basic data on most drug products approved since 1939—to obtain the approval date for each New Drug Application in our dataset.⁸⁵ In all, the patent and exclusivity information from every month between January 2005 and December 2015 amounted to 3834 pages of data that we sifted through by hand.

Drug strengths, in particular, posed data entry challenges. In the Orange Book, each strength of a drug is listed separately. Thus, if a certain patent or exclusivity applies to multiple strengths of a drug, the patent or exclusivity will be listed multiple times. In most cases, we found that if a patent or exclusivity was applied to one strength of a drug, it was eventually applied to all strengths of the drug. Thus, listing a patent or exclusivity multiple times in our dataset, for each corresponding strength, could amount to a form of double-counting and create an inaccurate picture of the level of patent and exclusivity activity. To choose the most conservative approach possible, we listed each patent and exclusivity that applied to a drug only once. This required extremely careful parsing of the Orange Book. In most cases, a list of added patents would be identical across all strengths of a drug, but occasionally, there were minute distinctions that could easily be missed, such as an extra patent added onto just one out of eight different strengths of the same drug.

More generally, when considering an analysis of how many drugs are involved in a particular behavior—in our case, how many drugs had patents or exclusivities added to them between 2005 and 2015—one must choose the level at which to conduct the analysis. The term ‘drug’ can have several different meanings, depending on the chosen definition and context. For example, one can choose to define a drug on the level of the active ingredient, the branded product name, the specific new drug application number, or the specific strength or formulation.

Consider the opioid addiction treatment drug, Suboxone. The active ingredients in Suboxone are buprenorphine hydrochloride and naloxone hydrochloride. There are, however, brand-name drug products other than Suboxone that are identified with the exact same two active ingredients, including Bunavail and Zubsolv. Moreover, within the brand-name Suboxone itself, there are two different new drug application numbers: drug application 20733, approved in October 2002, and drug application 22410, approved in August 2010. Within Suboxone drug application 22410 alone, there are four different strengths of the drug, corresponding to the same drug application number.

For our analysis, we chose to define ‘drug’ at the level of the new drug application number, given that many anecdotal reports indicate pharmaceutical game-playing at that level of granularity.⁸⁶ For example, if one version of a drug (at the new drug application level) is on the verge of losing patent protection, the pharmaceutical company might switch from a capsule to a tablet and submit a new drug application for the drug in tablet form, with new protections stemming from the revised formulation.⁸⁷ We did not go as far down as the level of strength, however, because we felt it could be misleading to define a 10-mg strength and a 20-mg strength of one drug as two separate drugs—resulting in counting two occurrences of strategic behavior—given the commonplace understanding of what ‘drug’ means. Moreover, as noted above, a patent or exclusivity applied to one strength was usually applied to all strengths of the drug.

There may, indeed, be game-playing involving different strengths of the same drug. For example, for a generic drug to receive approval, it must match the brand-name product in dosage strength.⁸⁸ If a new formulation does not have the same dosage or strength, pharmacists are not allowed to substitute the generic under most state drug substitution laws; such substitution is the major pathway for generic drug companies.⁸⁹ Thus, although we do not count the same patent applied to different dosages as more than one occurrence, our dataset does track instances in which a patent or exclusivity that had already been applied to one strength of a drug is applied to a new strength of that drug, so that future research can identify and analyse the behavior.

The definition of ‘drug’ could include drugs listed in Abbreviated New Drug Applications (ANDAs). ANDAs are the applications filed by companies seeking approval for a generic version of a drug.⁹⁰ Generic applications are likely to be listed in the patents and exclusivities section of the Orange Book; however, only in relation to what the Orange Book calls, the ‘PC’ or ‘patent challenge’ exclusivity, a 180-day period of exclusivity awarded to the first generic drug to successfully challenge a brand-name patent under Paragraph IV of the Hatch-Waxman Act. Our research, however, examines the use of exclusivities to obstruct generic entry. The 180-day exclusivity represents the exact opposite—the successful entry of a generic competitor—and thus, does not fall within the scope of our study. As such, we excluded all patent challenge exclusivities from our dataset and did not include generic drugs in our figures for the overall number of drug products.

There was also the question of which actions taken by the pharmaceutical companies should be considered part of the same game occurrence. If a drug has one patent added to it in March 2012 and one patent in April 2012, but both patents expire in April 2020, should we consider them to be part of the same game? It is certainly true that a larger number of patents have the potential to create greater barriers. Competitors wishing to challenge the validity of the protections built around a product in theory could be forced to overturn each and every one, although branded companies do not always choose to assert all of them. As a result, each addition does add to the arsenal of protection, increasing the difficulty of competitive entry.

Such multiple patents can be used for other strategic behaviors as well, even if they expire on the same date. Specifically, companies frequently separate their patent applications into different parts, which are then processed at the USPTO as what are known as continuations. Although they will all have the same final expiration date, they move through the patent office at different rates of speed and will be granted at different times. Having some pieces move more slowly allows companies to keep an eye on their competitors in the market, subtly adding language during the process that will better cover what a competitor has developed, although this may apply more appropriately in areas other than pharmaceuticals, which could have fewer direct competitors.⁹¹ The Federal Circuit has expressed its approval of this behavior.⁹²Nevertheless, a patent may be separated into different parts for perfectly legitimate reasons, and it is difficult at the level of data analysis we are applying to discern the difference with confidence. Thus, in the interest of fairness and careful conservatism, we did not count those as separate instances.

As a further exercise, however, we calculated our metrics in both scenarios—that is, counting patents that expire on the same day as separate and additions and not counting patents that expire on the same day as separate additions—to see the effect on our qualitative takeaways. We saw little qualitative difference between the two sets of calculations.

Finally, for the purposes of this paper, we included only small molecule drugs, rather than biologics, in our dataset. Small molecule drugs are simple, stable, single-molecule entities that are produced through chemical synthesis and are easy to replicate.⁹³ Commonplace drugs, such as aspirin, that are familiar to most people are small molecule drugs. In contrast, biologic drugs are large, complex products produced in living cell cultures for which it is currently impossible to create identical copies.⁹⁴ Examples of biologics include vaccines, blood products, and advanced gene therapies.⁹⁵ The FDA does not include biological products—or their generic counterparts termed ‘biosimilars’ or ‘interchangeables’—in the Orange Book but has established a separate publication, colorfully known as the ‘Purple Book’.⁹⁶

Unfortunately, the Purple Book is much less comprehensive than the Orange Book and does not include a patent and exclusivity section. As a result, our analysis could not extend to biologics. If data on the patents and exclusivities attached to biological products can be obtained in the future, whether through the FDA deciding to make such data public in the Purple Book or through a FOIA request, conducting an analogous inquiry into activity in the biologics sphere would be a worthwhile endeavor. Biologics and their generic counterparts are also a younger phenomenon. Congress created a system for expedited approval of copies of biologic drugs in 2010,⁹⁷ and the first biosimilar was approved only in 2015.⁹⁸ Thus, the skirmishes over generic versions of biologics are in their infancy. Over time, however, greater FDA reporting and transparency will be critical for tracking and evaluating behavior in this increasingly important sector of the industry.

III.B.iv. Transferring patent and exclusivity data from the 2005 annual Orange Book

The next step in assembling our dataset involved transferring over all patent and exclusivity information listed in the annual edition of the Orange Book from the year 2005 (as opposed to the cumulative supplements from 2005, which at this point, had already been entered into the dataset) in order to provide baseline information. Specifically, when a patent or exclusivity is marked as a new addition in a cumulative supplement, the Orange Book does not identify which component of the listing warranted the new addition flag. It could be that the entire listing—patent number, expiration date, patent codes, and all—is new, but it could also be that just one element is new. Thus, it was necessary to create baseline information to know which patents and exclusivities were already on the books at the start of our time period so that we could tease out which part of the listings flagged as new in any of the 2005 cumulative supplements constituted the addition. The annual edition for 2005 is published at the beginning of 2005, and it contains information that is current up to the last day of the previous year. Thus, entering the 2005 annual supplement provided the necessary baseline information for the initial year of our dataset.⁹⁹

III.B.v. Verifying the accuracy of the patent and exclusivity data

Ultimately, this process of collecting patent and exclusivity data for the 11 years from 2005 to 2015—both the monthly supplements and the 2005 annual edition—yielded 16,141 individual rows of data, with 9 to 11 data field columns per row. This amounts to over 160,000 individual cells of data, all entered by hand.

Any process of manually compiling over 160,000 individual data points, many of which were random strings of numbers, is subject to human error. Thus, after completing the dataset, we looked through the data second time and double-checked every entry from the monthly supplements and the 2005 annual edition for accuracy. A small number of errors were found and corrected.

We are optimistic that by double-checking every Orange Book listing in our dataset, we were able to catch the overwhelming majority of errors. Though it is certainly possible some errors remain, given the massive volume of data, we are confident that the overall conclusions would remain unchanged, even in the presence of a small number of data entry errors. Moreover, the coding process, which is described in the section below, effectively required us to go through the data line by line a third time, further reducing the possibility of significant inaccuracies in our dataset.

III.B.v.1.Coding the patent and exclusivity data.

As noted above,¹⁰⁰ Orange Book entries do not explicitly identify whether the entire listing is new or whether just one element of the listing is new, and if so, which component of the patent or exclusivity is new. In addition, the information that does exist requires careful interpretation. For example, in some cases, a patent listing appears identical to another previous listing. The only change is that while the patent was applied previously to strengths 1 and 2, for example, it is now being applied to strengths 3 and 4, as well. Although this might initially appear to be a new patent, to categorize it as such would be misleading, given that the substance of the change involves adding an existing patent to new strengths. These and many other circumstances necessitate individualized interpretation and analysis. Thus, we individually examined each line in our dataset, reading every entry in the context of the patents and exclusivities that came before.

The changes we tracked that we considered to be significant for our analysis of pharmaceutical game-playing included:

• Patents added for the first time, regardless of whether the addition included any drug substance, drug product, and/or use codes¹⁰¹;
• The addition of drug substance, drug product, and/or use codes to existing patents;
• Exclusivity additions (a full list of the exclusivities tracked can be found in Appendix A);
• Patents marked with a ‘delist request’ flag;¹⁰²
• Cases in which existing patents or exclusivities were added to a new strength of the same drug.
• There were other changes that we tracked but excluded from our analysis because it was unclear whether these changes were relevant to strategic pharmaceutical game-playing. These changes include cases in which:
• The patent term increased or decreased¹⁰³;
• A drug substance, drug product, and/or use code was removed;
• A change to a patent was applied to another use code listing of the same patent;¹⁰⁴
• A listing was determined to be an error in the Orange Book, whether made on the part of the company or the Orange Book staff—a category we call ‘errors’¹⁰⁵;
• A listing was determined to be a correction of a previous error on the part of the company or the Orange Book staff—a category we call ‘corrections’¹⁰⁶; and
• The Orange Book listing was ambiguous.¹⁰⁷

Some of the changes in the second list—changes we tracked but excluded from our analysis—could conceivably be related to pharmaceutical game-playing in one way or another. For example, there are cases in which a single drug product can receive multiple patent term extensions by strategically having two new drug applications approved on the same day and then extending a different patent for each.¹⁰⁸ Despite this possibility, our overarching philosophy in making methodological decisions was to err on the side of caution and make the conservative choice, with the result that, if anything, we are understating as opposed to overstating the results.

After completing the coding process, our data consisted of a complete set of every patent and exclusivity added to the Orange Book between January 2005 and December 2015, with each line categorized into a specific type of Orange Book addition or change. With this dataset in hand, we moved on to establishing a set of metrics for drawing conclusions from the large volume of data we had compiled and organized.

III.B.v.2. Establishing key metrics

As described above, our goal in assembling the dataset was to quantitatively evaluate the use of patents and exclusivities as a lifecycle management strategy for pharmaceutical products. To accomplish this task, we created metrics including the following:

• The number of drugs that had patents or exclusivities added to them in the Orange Book between 2005 and 2015, compared to the total number of drugs available between 2005 and 2015;
• The number of drugs that had patents or exclusivities added to them in the Orange Book, broken down by year for each year between 2005 and 2015;
• The number of drugs that had an exclusivity added to them, broken down by type of exclusivity;
• Exclusivities examined on this more granular level include orphan drug exclusivity, new patient population exclusivity, new product exclusivity, pediatric exclusivity, and indication exclusivity.¹⁰⁹
• The total quantities of patents and exclusivities added between 2005 and 2015;
• The number of drugs that had a high quantity of patents add to them in a single year between 2005 and 2015;
• The number of separate times that each drug had something added to it in the Orange Book (a measure of ‘serial offenders’);
• The number of drugs newly approved in a year compared to the number of drugs that had something added to them in the Orange Book in that year; and
• Percentage of the approximately 100 top-selling, non-biologic drugs between 2005 and 2015 that extended the initial ‘protection cliff’.

The Results section describes the metrics and their application, but the methodology of some metrics is best described here. Specifically, the first metric provides the total number of drugs that had a patent or exclusivity added to them, or had any other relevant change made in the Orange Book relative to the overall number of drugs in existence and listed in the Orange Book in the 11 years between 2005 and 2015.

The denominator in this metric—the overall number of drugs—required an immense amount of sleuthing through online data repositories and internet archiving sites to calculate with any level of precision. As with many other crucial pieces of FDA data, figures for the total number of drugs (at the level of new drug applications) listed in the Orange Book each year are not readily available.¹¹⁰ The FDA does make a copy of the Orange Book available in ASCII text, tilde-delimited format, which would be more easily imported to obtain an overall figure for the number of drugs with significantly less effort than hand-counting would require, but only for the most recent month.

Although the FDA currently updates the ASCII text file version of the Orange Book every month, this has not been the case across time. Internet archived versions show month-long periods that go by without a single change occurring in the ASCII text file versions of the Orange Book, while we know from the hard copy versions that dozens, or even hundreds, of changes occur each month.

On the flip side, the number of dates on which the archival system captured the FDA’s webpage and the distribution of those dates across any given year appear to be somewhat random. For example, for the year 2014, the webpage was captured once in February, once in April, twice in September, and three times in December. Meanwhile, in 2011, the webpage was captured every month of the year, at least two times each month. In September 2011, the number of days the webpage was captured reached a high of seven times, and there were a few occasions in 2011 that the webpage was captured more than one time in a single day. Thus, we compared each Internet-archived version of the Orange Book ASCII text files with the versions immediately before and after to cull out those archived versions that were mere duplicates.

Finally, we note that the comprehensiveness of our collection of Orange Book text files was at the mercy of whatever was available through Internet-archiving sources. It is possible that there was a gap between two of our archived webpages during which a certain drug was added and then removed. We would have no record of this drug’s existence in the Orange Book and consequently, it would not have been included in our count of unique drugs listed in the Orange Book between 2005 and 2015. This possibility is unlikely, however, given that there was rarely much of a temporal gap between the various versions we obtained. Moreover, most drugs would remain listed in the Orange Book for longer than the 1-week or 2-week periods for which we occasionally did not have any archived versions of the Orange Book.

With the archived versions in hand, we were able to obtain a figure for the total number of drugs (at the new drug application level) available in each year. We then combined the yearly information, sorting for unique new drug numbers among that aggregate list of new drug numbers, resulting in a figure for the total number of drugs available in our entire 2005–2015 timeframe. We compared the number of drugs that had patents, exclusivities, or other changes added on to them, between 2005 and 2015, to the total number of drugs available in those 11 years, to get a sense of how prevalent the behavior is in the overall universe of pharmaceutical products. The outcomes of this analysis will be detailed in the Results section.

Our final metric involves extension of what is commonly referred to as the ‘patent cliff’.¹¹¹ We examined the latest expiration date in the original set of protections and then determined if a new protection was subsequently added with a later expiration date.¹¹² We refer to this benchmark as the ‘protection cliff’ rather than the ‘patent cliff,’ given that many of the relevant ‘cliffs’ apparent in our dataset stemmed from exclusivities, not patents.

Our analysis focused on the best-selling drugs from the time period between 2005 and 2015, and, as with the entire study, we focused on non-biologic drugs. The high profit margins for blockbuster drugs provide a strong incentive for drug companies to invest in finding ways to extend protection. Thus, we chose the subset of our data for which we believed the protection cliff analysis would be most relevant.

To assemble a list of best-selling drugs from our study timeframe, we consulted the lists available through Drugs.com and Medscape.com. These websites obtain information from Verispan’s Vector One National (VONA) database and from the IMS Health database.¹¹³ From those lists, we selected the top 50, non-biologic brand drugs from each year.¹¹⁴ We then eliminated any duplicate drugs that overlapped in the top 50 from one year to the next. Our final grouping included a total of 106 best-selling drugs from the 10 years of 2005 to 2014, for which we analysed the frequency of protection cliff extension behavior.

We also chose to leave out the best-selling drugs from 2015. Our study only extends through 2015, and examining extension of a patent cliff requires a sufficiently long period of the drug’s lifecycle so that one can analyse movement across time. For drugs that did not have a first set of patents or exclusivities added to them until 2015, it would be impossible to analyse any future extension of the protection cliff.

One could argue that, on the whole, the later years in our dataset would be less fruitful for the same reason, thereby understating the results. This, of course, may be true and is consistent with our overall study design, which is intended to err on the side of understating results. In addition, with the later years in our dataset, there would at least be some possibility of relevant activity to analyse for those years, as opposed to 2016 for which there would be no possibility of examining any future extension of the protection cliff. Finally, the possibility that strategic behavior may be increasing over time makes the latest years important to consider.

IV. RESULTS

IV.A. Overview

The study results demonstrate definitively that the pharmaceutical industry has strayed far from the patent system’s intended design. The patent system is not functioning as a time-limited opportunity to garner a return, followed by open competition. Rather, companies throughout the industry seek and obtain repeated extensions of their competition-free zones. Moreover, the incidence of such behavior has steadily increased between 2005 and 2015, especially on the patent front and for certain highly valuable exclusivities. Most troubling, the data suggest that the current state of affairs is harming innovation in tangible ways. Rather than creating new medicines—sallying forth into new frontiers for the benefit of society—drug companies are focusing their time and effort extending the patent life of old products. This, of course, is not the innovation one would hope for. The greatest creativity at pharmaceutical companies should be in the lab, not in the legal department.¹¹⁵ The following sections describe the results obtained through our analysis in detail, but below are the key takeaways from the study:

• Rather than creating new medicines, pharmaceutical companies are recycling and repurposing old ones. In fact, 78% of the drugs associated with new patents in the FDA’s records were not new drugs coming on the market, but existing drugs. In some years, the percentage reached as high as 80%.
• Adding new patents and exclusivities to extend the protection cliff is particularly pronounced among blockbuster drugs. Of the roughly 100 best-selling drugs, more than 70% extended their protection at least once, with more than 50% extending the protection cliff more than once.
• Looking at the full group, almost 40% of all drugs available on the market created additional market barriers by having patents or exclusivities added to them.
• Many of the drugs adding to the Orange Book are ‘serial offenders’—returning to the well repeatedly for new patents and exclusivities. Of the drugs that had an addition to the Orange Book, 80% of those had an addition to the Orange Book on more than one occasion, and almost half of these drugs had additions to the Orange Book on four or more occasions.
• The number of drugs with a high quantity of added patents in a single year has substantially increased. For example, the number of drugs with three or more patents added to them in one year has doubled. Similarly, the number of drugs with five or more added patents has also doubled.
• Overall, the quantity of patents added to the Orange Book has more than doubled, increasing from 349 patents added in the year 2005 to 723 in 2015.
• The number of drugs that had a patent added to them in the Orange Book almost doubled.
• There were striking increases in certain exclusivities, such as orphan drug exclusivity, new patient population exclusivity, and new product exclusivity. In particular, the number of drugs with an added orphan drug exclusivity tripled. In addition, the number of times a use code was added to a patent more than tripled, suggesting that this has become a new favored game.

To provide a broad sense of the types of metrics we are using, some could be characterized as ‘intensity’ measures, which capture the breadth and depth of patent and exclusivity activity in the industry. Another set of our metrics can be characterized as ‘temporal’ measures, which evaluate whether there are any trends in the behavior under examination across time during our 11-year timeframe from 2005 to 2015.

IV.B. Number of drugs that had patents and/or exclusivities added to them in the Orange Book, compared to the total number of drugs available

As an initial inquiry, we wanted to determine the extent to which companies are adding patents and exclusivities to drugs. Is this a limited activity, confined to well-worn anecdotes that everyone repeats, or does it occur throughout the industry? Our results demonstrate that adding patents and exclusivities is a common behavior, endemic to pharmaceuticals. In fact, between 2005 and 2015, almost 40% of all drugs available on the market had patents, exclusivities, or other changes added to them.

Table 1 shows the total number of FDA-approved drugs available on the market in each year of our study. Table 2 shows the number of drugs that had a patent or exclusivity added to them as a percentage of the total number of drugs. The figure is broken down in terms of the number of drugs with an added patent, the number of drugs with an added exclusivity, and the number of drugs that had any relevant change made to it (which includes not only adding a patent and/or exclusivity, but also other significant changes such as adding a use code.)

IV.C. Increase in number of drugs with changes, broken down by year

To assess whether patent and exclusivity activity has undergone change over time or remained relatively stagnant, we broke down our data by year, looking first at the number of drugs with an added patent, then at the number of drugs with an added exclusivity, and then at the number of drugs with any relevant change made at all.

IV.C.i. Number of drugs that had a patent added to them, by year

As shown in Table 3, the number of drugs that had a patent steadily increased between 2005 and 2015, almost double from 166 drugs in 2005 to 300 drugs in 2015. This increase is also reflected in the percentage of drugs that with an added patent out of the total universe of drugs available in each year. While 6.91% of all drugs listed in 2005 added an added patent in 2005, 11.77% of all drugs listed in 2015 had a patent to them in 2015.

The upwards trend is even more apparent in visual form, as shown in Figure 1.¹¹⁶

IV.C.ii. Number of drugs that with an added exclusivity, by year

We also broke down the exclusivity data by year. This figure involved 19 different exclusivities, including well-known and highly significant ones, such as the orphan drug exclusivity and the pediatric exclusivity, but also lesser-known exclusivities, such as the GAIN (Generating Antibiotic Incentives Now) exclusivity.¹¹⁷

Unlike the patent data, the exclusivity data contained no discernable trend over time, as the numbers in Table 4 demonstrate. Given the number of exclusivities lumped together, however, any trends could be obscured by underlying trends—and perhaps opposing trends—within individual exclusivities. The graphic above contains a visual some readers may not be accustomed to. The gray shaded area around the blue prediction line is called a ‘prediction band.’ It represents the range of values we have 95% confidence will capture predictions and thereby provides a degree of reassurance in the validity of the results. Section IV D presents a more granular picture of individual exclusivities, identifying increases and decreases within the group, as different approaches gain and lose popularity.

IV.C.iii. Number of drugs with any relevant Orange Book change or addition, by year

In Table 5, we present figures for the number of drugs that made any relevant Orange Book change or addition, broken down by year for each year between 2005 and 2015. Such changes include not only adding a patent and/or exclusivity, but also other significant changes such as adding a use code.

As illustrated in Figure 2, there is a slight upward trend in the number of drugs with any relevant Orange Book change or addition, especially in the five most recent years between 2011 and 2015. It is unsurprising that the trend is subtle, given that this metric is largely a combination of the patent data, for which there was a well-defined upward trend, and the exclusivity data, for which there was no discernable trend.¹¹⁸

IV.D. Number of drugs with an added exclusivity, broken down by type of exclusivity

The lack of a trend line in the number of drugs with an added exclusivity over time could be due to the lack of trends in any of the individual exclusivities, but it could also be attributable to the cancelling out of opposing trends in individual exclusivities. To answer this question, we analysed the exclusivity data on a more granular level. By examining each of the 19 exclusivities included in our dataset individually, we found that there were several that exhibited increases in frequency between 2005 and 2015 and several others that exhibited decreases in frequency.

The exclusivities for which there was an upward trend include orphan drug exclusivity, new patient population exclusivity, new product exclusivity, and new use. The exclusivities for which there was a downward trend include pediatric exclusivity (both as applied to patents and to other exclusivities), and indication exclusivity. Below, we will focus on the two exclusivities that exhibited particularly strong increasing trends: The Orphan Drug exclusivity and adding a new use designation to an existing patent.

IV.D.i. Increase in orphan drug exclusivity

Orphan drug exclusivity is a 7-year exclusivity granted to drugs that are approved and designated specifically to treat diseases and conditions affecting populations of 200,000 individuals or fewer.¹¹⁹ The exclusivity was established through the Orphan Drug Act, originally passed in 1983 and amended through the Hatch-Waxman Act in 1984.¹²⁰ The orphan drug program was initially intended to spur investment in neglected fields of medical research and development—drugs to treat rare diseases that affect only a small number of people in the USA.¹²¹ Policy makers feared that there were insufficient financial incentives to develop treatments for small patient populations, and that as a result, these populations would languish untreated.¹²²Today, however, it seems that ‘everyone is an orphan’, with orphan drugs accounting for more than 40% of drugs approved by the FDA.¹²³

Part of the reason for the rapid expansion of the orphan drug program is the enormous value of the 7-year exclusivity. Most regulatory exclusivities awarded by the FDA extend a drug’s protected lifetime by a few months, or perhaps a few years at most. For instance, pediatric exclusivity extends exclusive marketing and data rights for a drug by 6 months, and the exclusivity awarded for new clinical studies lasts for 3 years.¹²⁴ At 7 years, orphan drug exclusivity is by far the longest lasting of the forms of regulatory property granted by the FDA. With such strong exclusivity protections, manufacturers of orphan drugs are able to raise prices to shockingly high levels. The median cost for a patient to use an orphan drug for a single year is nearly $100,000 dollars, compared to roughly $5000 for non-orphan drugs.¹²⁵ Given that just a few months of additional market protection can be worth hundreds of thousands of dollars for a drug company, winning an additional 7 years is akin to winning the lottery.

More important, drug companies have figured out how to raise prices under orphan drug protections, and then spread those high prices across patient populations much broader than the small groups envisioned at the passage of the Orphan Drug Act. This technique is referred to as ‘spillover pricing’.¹²⁶ The most common way that drug companies are able to accomplish spillover pricing is through off-label use, which occurs when doctors prescribe a medication for a use other than the one for which it was originally approved by the FDA.¹²⁷ Consider the drug Epogen, which was approved to treat a small population afflicted with anemia related to end-stage renal disease, and as such, received orphan drug designation.¹²⁸ After receiving this designation, however, Epogen was prescribed off-label to treat a wide variety of types of anemia, expanding the patient population paying the high price of Epogen dramatically.¹²⁹

Another approach for gaming the orphan drug exclusivity system is through ‘salami slicing’.¹³⁰ This strategy involves dividing up the patient population into separate slices—perhaps separating those with an early stage of the disease from those with an end stage, or those who developed a genetic disease from one mutation from those who developed it through another mutation—and obtaining a different orphan drug exclusivity for each slice. Through ‘salami slicing’, if the original and intended population for a drug is greater than 200,000, and thus too large to qualify for orphan drug designation, the drug company can simply divide that original group up into subpopulations that are small enough to qualify.

A drug does not actually have to be newly developed to qualify for orphan drug exclusivity. As such, long-existing drugs can be revived and repurposed for an orphan drug indication. In fact, a troubling investigation by one media organization concluded that one-third of orphan drugs approved since the program began in 1983 were either repurposed mass market drugs or drugs that received multiple orphan approvals.¹³¹

Consider the drug, 3,4-diaminopyridine (3,4-DAP), which was used by patients with a rare neuromuscular disease and had been shown to be safe and effective as early as 1983.¹³² Though the drug had never been officially approved, it had been provided to patients at no cost for many years thanks to a generous company and the FDA’s ‘compassionate use’ Investigational New Drug (‘IND’) program.¹³³ In 2015, however, a different company submitted an application for a slightly modified version of the drug that does not require refrigeration, obtaining orphan drug designation in the process.¹³⁴ As a result, the company projected that it would be able to charge between $37,500 and $100,000 per patient per year—for a drug that those same patients used to receive for free.¹³⁵

In our study, we found that the number of drugs with added orphan drug exclusivities to the Orange Book underwent a notable increase between 2005 and 2015, with a large jump between 2010 and 2011, and a steady climb upwards from 2011 through 2015 (see Figure 3).

Between 2005 and 2015, the number of drugs with an added orphan drug exclusivity tripled from 9 drugs in 2005 to 27 drugs in 2015. Between 2010 and 2015, the number of drugs with an added orphan drug exclusivity nearly quadrupled from 7 drugs in 2010 to 27 drugs in 2015.¹³⁶

IV.D.ii. Increase in new patient population exclusivity

The new patient population exclusivity is a subcategory within the ‘new clinical investigation’ exclusivity defined in 21 CFR 314.108. In categorizing exclusivities for the patent and exclusivity section of the Orange Book, the FDA has chosen to break down the new clinical investigation, more commonly known as new clinical studies, exclusivity into its constituent elements. Thus, the Orange Book does not contain a new clinical investigation exclusivity code, but it does contain codes for new patient population exclusivity, new product exclusivity, dosage schedule exclusivity, indication exclusivity, prescription to over-the-counter switch exclusivity, and a variety of other exclusivities that could stem from a new clinical study. Here we examine the new patient population exclusivity, which is a 3-year exclusivity granted to a drug that has been approved for use in a new patient population based on a new clinical investigation. For instance, the drug Seroquel (drug number 20639) received two periods of new patient population exclusivity: one for the treatment of schizophrenia in adolescents 13 to 17 years of age and one for the treatment of bipolar mania in children and adolescents 10 to 17 years of age.¹³⁷

Figure 4 shows the number of drugs with an added new patient population exclusivity for each year between 2005 and 2015.

For the new patient population exclusivity, there was a generally upward trend across time, though not as dramatic as that seen with orphan drug exclusivity. The number of drugs with an added new patient population exclusivity nearly tripled from 6 drugs in 2005 to 16 drugs in 2015. It is not altogether surprising that the trend in frequency of new patient population exclusivity mirrors that of orphan drug exclusivity, as the two exclusivities are intimately related. Orphan drug exclusivity is granted to drugs that are developed to treat small patient populations of fewer than 200,000 individuals. As discussed earlier,¹³⁸ drug companies often use ‘salami slicing’ to divide up a broader patient population into smaller, particularized populations that would qualify the drug for orphan drug exclusivity. Given that this technique often involves defining new patient populations, it makes sense that many drugs that qualify for orphan drug exclusivity might also qualify for new patient population exclusivity, and that an increase in orphan drug activity would correspond with a rise in grants of new patient population exclusivity.

IV.D.iii. Increase in new use codes

Our findings for the number of times a use code was added to a patent each year between 2005 and 2015 is shown in Figure 5,¹³⁹ and the number of drugs that had at least one use code added to them in each of those years is shown in Figure 6.¹⁴⁰

There was a notable increase in the number of use codes added to the Orange Book in our 11-year timeframe, rising from 115 use codes in 2005 to 364 in 2015. These results are corroborated by a study of use codes conducted by Kurt Karst, in which he found that the total number of use codes listed in the Orange Book nearly tripled between 2003 and 2013.¹⁴¹

The number of drugs with at least one added use code also exhibited an upward trend between 2005 and 2015, more than doubling from 63 drugs in 2005 to 173 drugs in 2015. One might attribute the rise in the number of drugs with at least one added use code to a general rise in the number of drugs with anything added to them to the Orange Book between 2005 and 2015. Even accounting for the rise in drugs with additions in the Orange Book, however, there is still a rise in the frequency of drugs with added use codes. In 2005, 63 out of the 233 drugs with any relevant addition or change to the Orange Book (27%) had at least one added use code. In 2015, 173 out of the 353 drugs that had any relevant addition or change made to them in the Orange Book (49%) had at least one added use code. Thus, the fraction of drugs with added use codes in the Orange Book rose from less than a third to just about one half during that 11-year period.

IV.E. Quantity of patents and exclusivities added between 2005 and 2015

An important distinction exists between the number of drugs that had a patent or exclusivity added to it and the total quantity of patents and exclusivities added. An individual drug could have just one added patent or one added exclusivity, but it also could have dozens of different added patents and exclusivities. Looking at total quantities of patents and exclusivities across the time period provides a picture of the amount of Orange Book activity at the level of sheer numbers of patents and exclusivities added, rather than at the level of the specific drugs tied to those patents and exclusivities. Similar to the previous metrics, we provide an aggregate figure for the entire time frame and then break down the numbers by year between 2005 and 2015. The results from this inquiry are shown in Table 6.

While there was no clear trend over time in the number of exclusivities added, there was a reliable increase across the 11 years in the number of patents added, especially in the last 5 years between 2011 and 2015. The quantity of patents added double from 349 patents in 2005 to 723 patents in 2015.

The increase over time for the quantity of patents added reflects the upward trend in the number of drugs with patent added to them each year between 2005 and 2015. Likewise, the lack of a trend in the quantity of exclusivities added in that time period corresponds with the absence of any pattern in the number of drugs that added an exclusivity across time.

The question is whether the increase in the quantity of patents is a sign of misbehavior on the part of the drug companies. For example, if more drugs were entering the market, we would see increase in the number of patents. This increase would likely be innocuous. On the other hand, if the number of patents per drug was increasing, there would be evidence of misbehavior as drug companies constructed broader and broader protections around existing drugs.

However, there is one scenario in which a static average number of patents could be a sign of an unhealthy patent system. If the average number of patents is high, the fact it is unchanging does not give it a clean bill of health. This, in fact, is a sign of entrenched habits of misbehavior. In that circumstance, we would not be seeing increasing misbehavior because misbehavior is the norm.

To determine which scenario the patent system is in, we analysed the average number of patents added per drug for each year between 2005 and 2015, shown in Figure 7.

Here we can see the system is in a combination of scenarios. The average number of patents added per year increased from an average of 1.7 patents in 2005 to an average of 2.25 patents in 2015. This increase is slight but non-negligible, and future researchers may wish to keep an eye on whether this continues.¹⁴² Moreover, the average number of patents added is high across all years.

These average figures are dragged down by the drugs that did not add any patents in a particular year, but the increase across our timeframe is still clearly evident. This indicates that the growth in the quantity of patents added between 2005 and 2015 is attributable to two factors working in concert: (1) the growth in the number of drugs adding patents and (2) the growth in the average number of patents added per each one of those drugs.

IV.F. Number of drugs that added a high quantity of patents in a single year

Our next metric examined the number of drugs that added a high quantity of patents in a single year. The growth in the average number of drugs per year could be due to many drugs adding a slightly higher number of patents or it could be due to a smaller subset of drugs adding a high quantity of patents. In Figures 8 and 9, we show the number of drugs that added a high quantity of patents in a single year, with a ‘high quantity’ defined as three or more patents in Figure 9 and five or more patents in Figure 10.

There was a clear increase in the number of drugs with three or more added patents in a single year between 2005 and 2015. The figure more than doubles from 37 drugs in 2005 to 76 drugs in 2015. When the definition of a ‘high quantity’ of patents was changed from three to five, the results were similar. The number of drugs with five or more added patents in a single year also doubled between 2005 and 2015, from 14 drugs in 2005 to 34 drugs in 2015.

The upward trend in the number of drugs with a high quantity of added patents in a single year seems to indicate that drug companies are increasingly applying for as many patents as possible and seeing what they get. Unfortunately, it is likely that as more patents are added to a drug, the quality of the patents declines. Typically, the subsequent patents are more likely to be ‘secondary patents’, which, instead of covering the active ingredient or base compound, cover modified forms of the active ingredient, associated uses of existing chemical compounds, new combinations of old chemical compounds, dosage regimens, and specific formulations (ie tablet vs. capsule).¹⁴³ For instance, while the first patent added to a drug might cover the core active ingredient of the drug, the fifth patent might be covering a therapeutically negligible change to the formulation or composition of the drug. As such, the increase in the number of drugs with a high quantity of added patents in a year might be an indication that the pharmaceutical game-playing strategy of evergreening is becoming increasingly common.¹⁴⁴

IV.G. Number of ‘serial offender’ drugs

Some drugs in our dataset had patent or exclusivities added to them in to the Orange Book only once during our 11-year timeframe. Other drugs, however, repeatedly returned to the well, having one set of patents and exclusivities added to them, then having another set added a few months later, coming back with another round a few years after that, and so on. To capture this behavior, we measured the number of months during which a drug had a patent or exclusivity added to it in the Orange Book. This means that regardless of whether the drug 1 patent or 10 patents added to it that month, we considered that month as one instance of patent activity. We did this to remain as conservative as possible in our calculations.

Table 7 shows that a surprisingly large percentage of drugs returned to the well repeatedly. Out of the drugs that had at least one Orange Book addition, 80% had additions on more than one occasion. Moreover, 49% received additions on four or more occasions, and 20% received additions on seven or more occasions. As these results demonstrate, drugs that repeatedly bolster their patent and exclusivity protections are not the rarity they might once have been.

IV.H. Number of drugs newly approved compared to number of drugs with to the orange book additions in a year

The next metric examined the number of drugs that were newly approved each year¹⁴⁵ compared to the number of drugs that had patents and exclusivities added to them in the Orange Book in each year. In other words, within all the drugs on the market that had patents and exclusivities added to them each year, which ones were drugs that were newly approved that year, and which ones were drugs that had been approved in the past.

This metric is significant in that it provides an indication of how much patent and exclusivity activity is due to innovation in the pharmaceutical industry and how much of it is attributable to the recycling or repurposing of old drugs. If the number of drugs with added patents and exclusivities to the Orange Book each year far exceeds the number of new drugs approved each year, the result suggests that many drugs are receiving patents and exclusivities—not for innovation represented by a newly approved drug—but rather for changes made to old drugs that were approved previously.

Of course, a company could bring a novel drug to market and not apply for any type of patent or exclusivity. It would be unlikely, however, for them to do so, given the associated market benefits of patents and exclusivities. A company could also gain approval for a new drug late in the year, and, in doing so, have those patents appear in the following year’s Orange Book. We could not eliminate that possibility from our dataset, which represents a limitation of our analysis. In addition, we would expect that the number of drugs falling into any year end would be small.

Finally, we should note that our analysis is likely to significantly understate the amount of repurposing and recycling of old drugs. We examined drugs at the New Drug Application level. Anecdotal evidence suggests that with some product-hopping and evergreening behavior, companies change the name and make insubstantial formulation changes to the drug, submitting the new product under a different ‘new drug application’ than the original one.¹⁴⁶ Our dataset did not connect different new drug applications to each other, and we could not capture that behavior. Thus, the dramatic results below are still only part of the dismal picture.

As is evident from Table 8, the number of drugs with additions to the Orange Book dwarfs the number of newly approved drugs in every single year between 2005 and 2015. On average, 77.62% of drugs with Orange Book additions in a particular year were not new approvals from that year. This suggests a large degree of repurposing and recycling of existing drugs in the pharmaceutical industry, and concomitantly, less innovation and invention than the patent process is intended to create.

When looking at just patent additions, the narrative is the same. As seen in Table 9, 77.55% of drugs with a patent added to them between 2005 and 2015 were existing drugs. Strikingly, this problem grew over the timeframe of our study, with the number of existing drugs with added patents almost doubling from 2005 to 2015.

The concern with repurposing and recycling of old drugs is the following: while many of the changes made to those old drugs may earn new patents and exclusivities, they may not be significant from a patient benefit or therapeutic point of view. As such, society may be lavishing expensive rewards on suboptimal behavior.¹⁴⁷ The concern is even greater if one considers that many of these secondary patents may be of questionable validity. This is not to suggest that the changes would never have any value, to any patient, under any circumstances. Rather, minor changes to dosage or delivery systems, for example, may have some amount of value to some patients. Similarly, applying an old drug to a different disease brings the advantage of years of experience in the field, which can provide important information on safety. These advantages may not, however, justify the magnitude of the patent reward that is conferred. From society’s standpoint, one might be better off with incentives that drive scientists back to the bench to look for advances of greater significance.¹⁴⁸

IV.I. Percentage of top 105 best-selling drugs from 2005 to 2015 that had the ‘protection cliff’ extended

Our final metric is the percentage of the 106 best-selling drugs between 2005 and 2014 that had the ‘protection cliff’ extended. Blockbuster drugs are the ones for which the pharmaceutical companies have the most to lose if their exclusivity period ends, and the most to gain by extending the lifetime of the drug, even by just a few months. Thus, if competition blocking behavior is to be found anywhere, it would be found here. The results from this metric—broken down between drugs that had their protection cliff extended at least once and drugs that had their protection cliff extended more than once—are shown in Table 10. The results are striking.

Out of the 106 top-selling drugs from between 2005 and 2014, more than 70% had their protection cliff extended at least once and more than 50% had their protection cliff extended more than once. The magnitude of the behavior highlights the extent to which stifling competition has become the norm in the pharmaceutical industry. When more than 70% of best-selling drugs had their protection extended, it is clearly the go-to approach for profitability.¹⁴⁹

One can easily anticipate such maneuvering to continue going forward, particularly given the top-selling drugs going off patent. Between 2014 and 2020, an estimated $253 billion in worldwide drug sales is at risk due to expiration of patents on blockbuster drugs.¹⁵⁰ Without societal action, the future is likely to look like more of the same.

V. SOLUTIONS

As described in the opening of this article, the intellectual property system in general and the patent system in particular are designed to provide an opportunity for innovators to garner a return. Competition may be held in abeyance for a limited time, but those who receive the benefit must pay for the privilege by disclosing sufficient information that competitors will be able to step in. This design reflects the deeply rooted notion that providing a period of exclusivity for inventors is intended to rebound to the benefit of society as a whole, not simply to the benefit of the inventors. The patent protection should end, returning the market to a competitive state.

This foundational structure of the patent system—one that delicately balances innovation and competition—is crumbling, whittled away across time as one good idea after another creates a special carve-out. Each carve-out, standing on its own, presents an appealing cause. Together, however, the result is a complete undermining of the system for pharmaceutical innovation as the repeated addition of protections, one after another, pushes competition further into the future, threatening innovation in the process. The behavior is not limited to a few bad apples. Our research reveals that it is endemic to the pharmaceutical industry.

In short, this is not an image of innovation and competitive entry. It is an image of a system that provides for repeated creation of competition-free zones, pushing a competitive market further and further out into the future. The problem is not only pervasive and persistent, but it is also growing across time.

The impact created by these repeated competition zones is not some abstract problem that our grandchildren may face. Rather, the nation’s pharmaceutical system is in crisis today, with prices soaring to heights that distort both individual and government budgets.¹⁵¹ These dire circumstances bring calls for price controls, for government marching in to direct drug production, and for other strong measures.¹⁵² The US Government’s history of directly managing pharmaceutical innovation, however, has been disappointing. In fact, prior to the Bayh-Dole Act of 1980, the federal government took responsibility for handing out licenses for innovation developed through government-funded research. Bayh-Dole shifted that responsibility from the federal government to universities, precisely because the government failed so miserably in this role. There is little reason to expect a different result this time.¹⁵³

Competition is a powerful and effective tool, however, and paving the way for competition whenever it is possible remains the optimal approach. When the government itself bestows benefits that are stifling competition, society has both an obligation and an opportunity to act. One cannot, however, enter into such action lightly; it must be designed with thought and care. Pharmaceutical research and development are expensive, and companies must have sufficient incentive to travel down that risky road. Nevertheless, by incentivizing game-playing rather than innovation, society has clearly missed the mark.

V.A. One-and-done

This study offers a disappointing view of the state of pharmaceutical innovation, but this result is not inevitable. With sufficient political will—always a challenging task in the US landscape—our valued patent system can operate in the manner intended. The following section sketches out an approach that could roll back the repeated creation of competition-free zones documented in our research.

Specifically, one could implement of the principle of ‘one-and-done’ in which a drug would receive one period of exclusivity, and only one. The choice of which ‘one’ could be left entirely in the hands of the pharmaceutical company, with the election made at the moment of drug approval. Perhaps development and approval on the drug has gone swiftly and smoothly, so the remaining life of one of the drug’s patent is of greatest value. Perhaps those processes languished through many setbacks, such that designation as an orphan drug or some other benefit would bring greater reward. The choice would be up to the company itself, based on its own calculation of the maximum benefit. The result, however, would be that a pharmaceutical company must choose whether its period of exclusivity should be a patent, or an orphan drug designation, or a period of data exclusivity for safety and efficacy data, or something else—just not all of the above and more.

Crafting the one-and-done implementation at the FDA level underscores the fact that these problems and solutions are designed for pharmaceuticals, not other types of technologies. Although there are similarities within the patent system for all inventions, given the drug approval processes—including the Hatch-Waxman system for approval of generic drugs and the Biosimilars pathway for approval of follow-on biologics—pharmaceuticals are different.

Much of one-and-done could be implemented through legislative changes to the FDA drug approval system, which would apply to patents granted going forward. Statutory amendments could specify that once a company elects a particular patent or exclusivity, competitors wishing to obtain approval of a generic version of the drug through the Hatch-Waxman system need only certify to that one exclusivity.

The election could be crafted so that it mandates relinquishment of any other patent or exclusivity claims as to the generic drug being approved. This approach would be somewhat analogous to an election that currently exists under the current Hatch-Waxman Act. When a generic applicant makes a Paragraph IV certification claiming that the brand-name company’s patents are invalid or do not apply to the drug, the brand-name company has a period of time to challenge that certification in court. If the brand-name company fails to challenge the assertion, it relinquishes various rights, particularly, the right to an automatic 30-month stay of the generic’s approval. Without the 30-month stay, the brand would have to prove likelihood of success on the merits and other preliminary injunction factors to keep the generic off the market during the period of the litigation.¹⁵⁴ Similarly, in the proposed system, the company’s choice to designate a particular form of exclusivity upon approval could serve to relinquish its right to challenge the generic under any other exclusivity.

The compromise embodied in the original Hatch-Waxman system could provide a model for crafting the one-and-done system. Specifically, the Hatch-Waxman legislation included both the expedited system for generic approval and an expansion of the patent term to reflect delays in the federal approval system. Similarly, in implementing a one-and-done system, Congress could choose to expand the periods of the current exclusivities available, in recognition that pharmaceutical companies will no longer be able to tack so many periods of protection on to each other. One could argue that the current effects of the patent and exclusivities were never what Congress intended. Nevertheless, it may be politically expedient to follow the path blazed by Hatch-Waxman. In a similar vein, Congress could choose to standardize the periods of protection offered by various exclusivities, which currently range from 6 months to 7 years. As described in the section below on simplification, the complexity of these various systems provides opportunities for game-playing. Standardization may reduce those opportunities.

Some commentators may be tempted to claim that any relinquishment of patent or exclusivity rights constitutes a taking of private property. In particular, one scholar, Adam Mossoff, has asserted that patent rights are constitutionally protected property, and as such, would be subject to the Fifth Amendment Takings Clause.¹⁵⁵Even Mossoff, however, acknowledges that ‘modern courts and scholars … seem to agree in a rare case of unanimity that the historic record reflects no instance of a federal court holding that the Takings Clause applies to patents’.¹⁵⁶

In a 2018 case upholding the inter partes review system at the Patent and Trademark Office, the Supreme Court specifically avoided ruling on the question of whether patents are property for the purpose of the Takings Clause.¹⁵⁷ In paragraph that begins by noting, ‘[w]e emphasize the narrowness of our holding’ and presents a litany of what the decision does not address, the court cryptically concludes by noting that ‘our decision should not be misconstrued as suggesting that patents are not property for purposes of the Due Process Clause or the Takings Clause’. The Justices then cite two cases related to sovereign immunity and whether the state can be sued for using a patented item without paying for it.¹⁵⁸ In contrast, the two dissenting Justices use language that would move the status of patents much closer into the realm of traditional property rights. This suggests that whether, and the extent to which, patents are treated as property for Constitutional purposes is likely to arise in future Supreme Court cases.

The notion of patents as full property rights—akin to the type of core property rights protected by the Constitution—would require ignoring significant aspects of patent history and theory. Patent rights are theoretically, doctrinally, and practically distinct from real property, making the notion of an absolute right to exclude particularly inapplicable¹⁵⁹. More important, unlike real property, patent rights are granted by the government for limited times and for limited purposes, namely promoting the progress of the useful arts¹⁶⁰ for the benefit of society. The utilitarian roots of their theory and design bear little resemblance to natural rights theories of the types of property protected by the Constitution.

One particularly cogent modern description of the issue appears in a dissent to the 2015 Supreme Court decision in the Teva case, in which the dissenters reviewed the history of patent rights in contrast to core property rights.

The Anglo-American legal tradition has long distinguished between ‘core’ private rights—including the traditional property rights represented by deeds—and other types of rights. These other rights [include] ‘privileges’or’franchises,’ ‘which public authorities have created purely for reasons of public policy and which ha[ve] no counterpart in the Lockean state of nature. Notwithstanding a movement to recognize a core property right in inventions, the English common law placed patents squarely in the final category as franchises.¹⁶¹

As the text of the dissent also explained, our own ‘Framers adopted a similar scheme’.¹⁶²

In short, patents are not core property rights, and attempting to characterize them as such threatens the reverence that the nation has traditionally held for core property rights. Although concerns may play out in the context of due process, the notion that the Takings Clause would prevent Congress from shortening the length of time or the interaction among various patent rights¹⁶³ would be misguided, at best.

V.B. Ruthless simplification

For those who like complexity, the intellectual property system for pharmaceuticals is a garden of delights. From the Hatch-Waxman legislation, to the Biosimilars Act, to the maze of regulatory exclusivities, and beyond, the judicial and regulatory processes surrounding intellectual property rights for drugs constitute among the most complex corners of our legal system.

Of course, some complexity in pharmaceuticals is inevitable. The intellectual property systems for drugs must, of necessity, interact with approval processes, and those approval processes must operate with exquisite awareness of public health and safety. These are heady responsibilities. Nevertheless, the system has become so complex and convoluted that it threatens to collapse in on itself.

And, of course, complexity breeds endless opportunities.¹⁶⁴ It ensures that the legislators and regulators will always be at least a step behind in an endless game of cat and mouse. Year after year, government actors must attempt to block strategic behaviors that have developed, even as the industry develops new ones.¹⁶⁵ In such a process, it is clear that our incentive structure is badly misaligned with societal goals.

Putting the system back on track will require ruthless simplification.¹⁶⁶ It means stripping away the intricate details that are so appealing to those who must form compromise among interest groups, but that sow the seeds of current and future strategic behavior. In short, what has become business as usual for the pharmaceutical industry must become a thing of the past.

The 180 day period of exclusivity for first-filing generics is a classic example of complexity that provides game-playing opportunities. It is an extremely complicated and intricate piece of legislation. In what is known as pay-for-delay, the system has provided a method for generics and branded pharmaceutical companies to form anticompetitive agreements in which the generic agrees to stay off the market in exchange for some form of payment.¹⁶⁷ A simplified approach, in which the period of exclusivity attaches only if the patent is actually invalidated, could reduce the game-playing. Frankly, although it is akin to heresy to suggest, one could argue that the entire first filer exclusivity period should be eliminated. There may be sufficient market opportunities for generics without that incentive, particularly given the high price of branded pharmaceuticals, and the game-playing it has spawned may outweigh the benefits of having such as system.

V.C. Transparency¹⁶⁸

Systemic changes such as One-and-Done and Ruthless Simplification require both time and political courage to promulgate and implement. Thus, additional adjustments will be necessary along the way. Chief among these is transparency. As one commentator noted in a 2017 FDA public meeting, transparency ‘is the enemy of all this abuse’.¹⁶⁹ The power of pharmaceutical company behavior lies, in part, from the obscurity of those behaviors, making it difficult for legislators, regulators, and the public to tease out and address what is happening. No matter which approaches are chosen for addressing improper behavior in the pharmaceutical realm, transparency will be a critical component. Only then will society be able to quickly identify new strategic behaviors as they emerge and provide solutions before too much damage occurs.

Moreover, competition thrives on information. Those willing to offer better terms will find willing buyers—from the federal government to private insurers to HMOs—beating a path to their door and driving some measure of competition into a non-competitive market. That competitive environment can only exist if potential competitors have full and complete information. With this in mind, the following sections provide examples of areas in need of increased transparency. Although we focus these suggestions on transparency in relation to the topics studied in this article, we note that transparency in drug pricing will be critical as well, in order to take in the full range of modern games.¹⁷⁰

V.C.i. Accessibility of Orange Book information

Accessibility of information is a problem throughout the FDA’s many resources. We faced a number of obstacles conducting research, which are briefly outlined here, although they are detailed more extensively in our other publications and in sections above.¹⁷¹

First, there is considerable room for improvement within the Orange Book system, particularly for the information regarding ANDAs. One of the largest difficulties in that realm lies in locating the date of an ANDA’s filing, information that is invaluable to many researchers and members of the interested public.¹⁷² For our research, we had to painstakingly read through every published approval letter to pull the dates of original filings, which were often casually mentioned.¹⁷³ For many approved drugs, we had to estimate the quarter year in which the application was published based on the number sequence, a complex and time-consuming process. Even worse, there are recent reports that approval letters will no longer include the original date of filing, making an approach like ours almost impossible for drugs going forward.¹⁷⁴ In addition, the citizen petition files do not always link to specific generic applications or offer indications of this information that is so critical for tracking the timing of citizen petitions in relation to the application process for a particular drug.¹⁷⁵

Prior editions of the Orange Book are not readily available, although the FDA has stated that they will eventually be available in an archive.¹⁷⁶ There is also a lack of clarity when information is updated or changed. Specifically, when a patent or exclusivity is marked as a new addition in a cumulative supplement, the Orange Book does not identify which component of the listing warranted the new addition flag. It could be that the entire listing—patent number, expiration date, patent codes, and all—is new, but it could also be that just one element is new.

Timing issues exist with the updates to the Orange Book as well. While the FDA has begun updating the ASCII file version monthly, in the past it was not updated for months, even though hundreds of changes were occurring based on archived hard copies.¹⁷⁷ The information is difficult to interpret, especially in older versions. For example, files prior to December 11, 2009 do not include a data field indicating whether a drug product is a generic or a new drug.

In short, the Agency publishes substantial information online and in hardcopy, but a significant amount of information is missing.¹⁷⁸ This inaccessibility obscures the strategic behavior that is occurring, making it difficult for regulators, legislators, and the public to identify and address improper activity.¹⁷⁹

V.C.ii. Accessibility of information in the Purple Book

Our study examined behavior in the market for small molecule drugs through the lens of information in the Orange Book. In 2010, Congress approved a separate system for approval of follow-on biologic drugs, known as the Biosimilars Act. Information on biosimilars is listed in the so-called Purple Book, in colorful contrast to the Orange Book. While the Orange Book has long been a highly useful tool for various stakeholders addressing many aspects of small molecule drugs and their approved counterpart generic drugs,¹⁸⁰ the relatively new Purple Book was intended to be its equivalent for biologics, listing biologics and corresponding licensed biosimilars.¹⁸¹ Unfortunately, the Purple Book has serious shortcomings as an information vehicle. Although these shortcomings may be driven by the fact that the Biologics Act does not disclose particular information for the purposes of the judicial and regulatory challenges involved in the Act, the FDA could, of course, choose to provide the information, and Congress could mandate it.

For example, the Purple Book does not include patent information for the reference biological product.¹⁸² Although the Biologics Act does not require that a drug company publishes patent information up front,¹⁸³ that information may emerge to some extent during the process of approving what are known as biosimilars and interchangeables. As of now, it appears that the FDA has no plans to include patent information as it emerges in the biosimilar approval process.¹⁸⁴

Most important, the information available in the Purple Book is difficult to access. While the FDA has an easy-to-use, reasonably sophisticated website for the Orange Book—where a user can search by active ingredient, proprietary name, patent, applicant holder, or application number—in most cases there is no similar mechanism for the Purple Book.¹⁸⁵ In fact, the Purple Book’s two lists are only available in PDF format and are not easily searchable.¹⁸⁶

Biosimilars have extraordinary potential to lower pharmaceutical costs and expand access for consumers. If the FDA wishes to allow companies, academics, and other stakeholders to tap into this potential, the Purple Book must be updated to increase the amount of information available and to improve the accessibility of this information. At the very least, the Purple Book should be of the same caliber as the Orange Book; and it should aspire to even better.

V.D. Paving a better road

One-and-Done and Ruthless Simplification, coupled with transparency measures, could go a long way towards returning the system of pharmaceutical innovation to its proper competitive pathway. There will, of course, be much wailing and gnashing of teeth. The pharmaceutical industry has become comfortably accustomed to working with a system that provides space for creating non-competitive environments. The industry will not relinquish this environment with ease and grace, and the nation is likely to hear impassioned pleading that pharmaceuticals cannot withstand any reform of the current system.¹⁸⁷ Along similar lines, the CEO of the pharmaceutical company Allergan published a 2017 Op-Ed in the Wall Street Journal arguing that the 2011 patent reforms, which created a new post-grant review process for patents, left the company with no choice but to transfer their patents to Indian tribes to avoid having the patents reviewed.

When companies plead with government for benefits by arguing that they cannot withstand competition, one should be deeply skeptical. Our challenge as a society is to restore the balance provided by the patent system itself, in which the inventor of a truly innovative product receives a limited period time to attempt to garner a return, following which, open competition reigns supreme. The system has strayed far from that ideal.

Footnotes

ACKNOWLEDGEMENTS

Research for this piece was funded in part by a generous grant from the Laura and John Arnold Foundation. I wish to thank Mark Lemley, and Kurt Karst for their insights and comments. I am grateful beyond measure to Giora Ashkenazi, John Gray, Prianka Misra, Rabiah Oral, and Nick Thieme for research assistance. Finally, I wish to express my deep gratitude to Connie Wang, without whom the project would not have been possible, for her extraordinary contributions to research design and implementation.